The Transcription Factor 7-Like 2-Peroxisome Proliferator-Activated Receptor Gamma Coactivator-1 Alpha Axis Connects Mitochondrial Biogenesis and Metabolic Shift with Stem Cell Commitment to Hepatic Differentiation

(1)

RESEARCH OUTPUTS / RÉSULTATS DE RECHERCHE

Author(s) - Auteur(s) :

Publication date - Date de publication :

Permanent link - Permalien :

Rights / License - Licence de droit d’auteur :

Institutional Repository - Research Portal

Dépôt Institutionnel - Portail de la Recherche

researchportal.unamur.be

University of Namur

The Transcription Factor 7-Like 2-Peroxisome Proliferator-Activated Receptor Gamma

Coactivator-1 Alpha Axis Connects Mitochondrial Biogenesis and Metabolic Shift with

Stem Cell Commitment to Hepatic Differentiation

Wanet, Anaïs; Caruso, Marino; Domelevo Entfellner, Jean Baka; Najar, Mehdi; Fattaccioli,

Antoine; Demazy, Catherine; Evraerts, Jonathan; El-Kehdy, Hoda; Pourcher, Guillaume;

Sokal, Etienne; Arnould, Thierry; Tiffin, Nicki; Najimi, Mustapha; Renard, Patricia

Published in:

Stem Cells

DOI:

10.1002/stem.2688

Publication date:

2017

Document Version

Publisher's PDF, also known as Version of record

Link to publication

Citation for pulished version (HARVARD):

Wanet, A, Caruso, M, Domelevo Entfellner, JB, Najar, M, Fattaccioli, A, Demazy, C, Evraerts, J, El-Kehdy, H,

Pourcher, G, Sokal, E, Arnould, T, Tiffin, N, Najimi, M & Renard, P 2017, 'The Transcription Factor 7-Like

2-Peroxisome Proliferator-Activated Receptor Gamma Coactivator-1 Alpha Axis Connects Mitochondrial

Biogenesis and Metabolic Shift with Stem Cell Commitment to Hepatic Differentiation', Stem Cells, vol. 35, no.

10, 35(10), pp. 2184-2197. https://doi.org/10.1002/stem.2688

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal ?

Take down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

(2)

The Transcription Factor 7-Like 2–Peroxisome

Proliferator-Activated Receptor Gamma

Coactivator-1 Alpha Axis Connects Mitochondrial

,

d

P

ATRICIA

R

ENARD a

Key Words. Hepatic differentiation•Stem cells•Mitochondria•Oxidative phosphorylation•

Wnt/b-catenin

A

BSTRACT

Increasing evidence supports that modifications in the mitochondrial content, oxidative phos-phorylation (OXPHOS) activity, and cell metabolism influence the fate of stem cells. However, the regulators involved in the crosstalk between mitochondria and stem cell fate remains poorly characterized. Here, we identified a transcriptional regulatory axis, composed of transcription factor 7-like 2 (TCF7L2) (a downstream effector of the Wnt/b-catenin pathway, repressed during differentiation) and peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC-1a) (the master regulator of mitochondrial biogenesis, induced during differentiation), coupling the loss of pluripotency and early commitment to differentiation, to the initiation of mitochon-drial biogenesis and metabolic shift toward OXPHOS. PGC-1a induction during differentiation is required for both mitochondrial biogenesis and commitment to the hepatocytic lineage, and TCF7L2 repression is sufficient to increase PGC-1a expression, mitochondrial biogenesis and OXPHOS activity. We further demonstrate that OXPHOS activity is required for the differentia-tion toward the hepatocytic lineage, thus providing evidence that bi-direcdifferentia-tional interacdifferentia-tions control stem cell differentiation and mitochondrial abundance and activity. STEM CELLS

2017;35:2184–2197

S

IGNIFICANCE

S

TATEMENT

Mitochondria, which represent the major source of energy and power for the cells, are increas-ingly recognized as major regulators of stem cells. In this study, we show that increasing the content and activity of mitochondria is essential for stem cell differentiation into hepatocytes. We identified a regulatory axis that couples the loss of pluripotency and commitment to differ-entiation to the increase of mitochondria abundance and activity. In turn, mitochondrial activity further supports the differentiation process. By better deciphering the interplay between mito-chondria and stem cell differentiation, these findings could benefit the development of stem-cell based therapeutics and regenerative medicine.

I

NTRODUCTION

Mitochondrial biogenesis and metabolism recently emerged as critical modulators of stem-ness properties and differentiation programs [1]. Inhibiting mitochondrial biogenesis or function prevents or disturbs the differentiation of stem cells into multiple lineages [2] and facilitates the reprogramming of somatic cells into induced pluripotent stem cells [3]. These

data support a crucial role for mitochondria and energy metabolism in the regulation of plu-ripotency and differentiation-related pathways. However, the molecular regulators involved in this crosstalk between pluripotency and differ-entiation on the one hand, and mitochondrial biogenesis and metabolism on the other hand, are still poorly characterized.

Conceptually, one can distinguish upstream and downstream regulators of the interplay

a

Laboratory of Biochemistry and Cell Biology (URBC), NAmur Research Institute for LIfe Sciences (NARILIS), University of Namur (UNamur), Namur, Belgium;

b

South African National Bioinformatics Institute (SANBI)/Medical Research Council of South Africa Bioinformatics Unit, University of the Western Cape, Bellville, South Africa;

c

Laboratory of Clinical Cell Therapy, Institut Jules Bordet, Universite Libre de Bruxelles (ULB), Brussels, Belgium;

d

Laboratory of Pediatric Hepatology and Cell Therapy, Universite Catholique de Louvain, Institut de Recherche Clinique et Experimentale (IREC), Brussels, Belgium;

e

Department of Digestive Diseases, Institut Mutualiste Montsouris, Paris Descartes University, Paris, France

Correspondence: Patricia Renard, Ph.D., Laboratory of Biochemistry and Cell Biology (URBC), University of Namur, rue de Bruxelles 61, 5000 Namur, Belgium. Telephone: 1320-81-724-128; Fax: 1320-81-724-135;

e-mail: [email protected] Received September 18, 2017; accepted for publication July 15, 2017; ﬁrst published online in STEMCELLSEXPRESSAugust 10, 2017.

http://dx.doi.org/ 10.1002/stem.2688

(3)

between mitochondrial biogenesis and stem cell differentiation. The upstream regulators of the interplay could be defined as the regulators involved in pluripotency and/or differentiation pathways responsible for the control of mitochondrial biogene-sis and metabolism. The downstream regulators on the other hand, could be defined as the regulators of cell differentiation that are activated following the induction of mitochondrial biogenesis and metabolic shift. An example of upstream regulator is the transcription factor hypoxia inducible factor 1 alpha (HIF-1a). In embryonic stem cells (ESCs), HIF-1a positively regulates the expression of pluripotency genes while inhibiting the expression of peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC-1a) [4], the master regulator of mitochondrial biogenesis [5, 6]. Therefore, HIF-1a maintains pluripotency while inhibiting mitochondrial biogenesis in ESCs [4]. An example of downstream effectors are the mitochondrial reactive oxygen species (ROS) produced during the adipogenic differentiation of bone marrow mesenchymal stem cells (BM-MSCs), which are required for the induction of the PPARc, a master regulator of adipogenesis [7]. The question whether those regulators are involved in the crosstalk between mito-chondrial biogenesis and stem cell differentiation in different models remains to be determined. It is likely that stem cell and lineage-specificities exist, and that other molecular actors may act together and/or independently of them in that process.

In this study, we investigated the crosstalk between mito-chondrial biogenesis and the hepatocytic differentiation of BM-MSCs. We found that the mitochondrial biogenesis and meta-bolic shift toward oxidative phosphorylation (OXPHOS) occur within the ﬁrst days of hepatocytic differentiation and overlap the loss of genes belonging to multipotency-related pathways. Bioinformatics analyses of transcriptomics data collected during the early steps of the differentiation process enabled us to iden-tify multiple putative co-regulators of mitochondrial biogenesis and cell differentiation, among which is PGC-1a. We found that PGC-1a is required both for the induction of mitochondrial bio-genesis and for the proper differentiation of BM-MSCs into hepatocyte-like cells. To understand how PGC-1a expression is induced during hepatocytic differentiation, we analyzed the expression of predicted regulators of PGC-1a. Among them is transcription factor 7-like 2 (TCF7L2), a downstream effector of the Wnt/b-catenin signaling pathway. We evidenced that TCF7L2 controls PGC-1a expression, mitochondrial biogenesis, and OXPHOS activity in BM-MSCs, supporting the hypothesis that TCF7L2 repression during hepatocytic differentiation initiates PGC-1a induction, mitochondrial biogenesis, and meta-bolic shift.

M

ATERIALS AND

M

ETHODS

Cell Culture and Differentiation

Cultures of human BM-MSCs were established as described pre-viously [8] after informed consent from healthy donors and the approval of the local ethics committee of the Jules Bordet Insti-tute (Belgium). For all studies, at least three independent repli-cates using cells isolated from different donors were performed. All experiments were performed using cells from passages 4–7.

BM-MSCs were expanded in 1 g/l glucose Dulbecco’s Modi-ﬁed Eagle Medium supplemented with 1% penicillin–streptomy-cin (PS) mixture (Life Technologies, Carlsbad, CA, now part of

Thermo Fisher (https://www.thermofisher.com/us/en/home/ brands/life-technologies.html)) and 10% fetal bovine serum (FBS, PAA/GE Healthcare, Chicago, IL, http://www3.gehealthcare.com) (“Exp” condition). Before the hepatocytic induction, BM-MSCs were seeded on type-1 collagen (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com)-coated dishes and expanded till 95% confluency (“day 0”) (illustrated on Fig. 2). Hepatocytic induction was carried out by incubating BM-MSCs in Iscove’s Modified Dulbecco’s Medium (IMDM, Life Technologies) contain-ing 1% PS mixture and supplemented with 20 ng/ml epidermal growth factor (PeproTech, Rocky Hill, NJ, https://www.peprotech. com) and 10 ng/ml basic-fibroblast growth factor (FGF-2, Pepro-Tech) for 48 hours and then with 10 ng/ml FGF-2, 20 ng/ml hepa-tocyte growth factor (PeproTech), 0.61 mg/ml nicotinamide (Sigma-Aldrich, Saint-Louis, MO, https://www.sigmaaldrich.com), and insulin–transferrin–selenium (ITS) mixture 1X (Life Technolo-gies) for 10 days. Cells were then treated with a hepatocytic mat-uration cocktail containing 20 ng/ml oncostatin M (PeproTech), 1 lM dexamethasone (Sigma-Aldrich) and ITS for 10 days. Undif-ferentiated BM-MSCs were maintained in IMDM supplemented with 1% FBS and 1% PS. For experiments involving OXPHOS inhibitors, cells were differentiated or left undifferentiated in the presence of 1lM myxothiazol (Sigma-Aldrich), 5 lM antimycin A (Sigma-Aldrich), 1lM oligomycin (Sigma-Aldrich), or the corre-sponding vehicles (1:2,000 dimethyl sulfoxide for myxothiazol and 1:2,000 ethanol for antimycin A and oligomycin) throughout the differentiation time course.

Lentiviral Vectors Production and Lentiviral

Transduction

Lentiviral vectors were produced in HEK293T cells following transfection of envelope plasmid (pCMV-VSV-G, Addgene, Cam-bridge, MA, https://www.addgene.org, #8454), packaging plas-mid (psPAX2, Addgene #12260), and expression plasplas-mid [pLKO.1 puro backbone containing the non mammalian shRNA control (shNT, SHC002) or shRNA against PGC-1a (TRCN0000001165) or TCF7L2 (TRCN0000061897) (Sigma-Aldrich)]. Lentiviral vectors were titrated using the quantitative polymerase chain reaction (qPCR) Lentivirus Titration kit (Applied Biological Material, Van-couver, Canada, https://www.abmgood.com), according to the manufacturer’s instructions. BM-MSCs at 60% conﬂuency were transduced in 25 cm2ﬂasks with 203 106millions of lentiviral particles in the presence of 60 lg/ml protamine sulphate (Sigma-Aldrich). Transduced cells were selected with puromycin 1lg/ml for 4–6 days and left without puromycin at least 2 days before experiments.

Massive Analysis of cDNA Ends Analysis

Total RNA was isolated using the RNeasy Mini kit and Qiacube (Qia-gen, Hilden, Germany, https://www.qiagen.com). For each condi-tion, equal amounts of RNA isolated from ﬁve independent donor cell batches were pooled and submitted to massive analysis of cDNA ends (GenXPro, Frankfurt, Germany, http://genxpro.net). Transcript per million normalization and tests for differential expression of mapped reads were performed in the statistical pro-gramming language R (www.rproject.org/) using the DEGexp func-tion of the DEGseq package [9], after replacing the “0” counts by the arbitrary “0.5” value. Data were deposited in NCBI’s Gene Expression Omnibus [10] and are accessible through GEO Series accession number GSE75184. Genes were considered as differen-tially expressed if upregulated or downregulated with a p< .05 or

(4)

with a p< .25 concomitantly with a |log2 fold-change| > 0.3785 (i.e., with fold changes>1.3 or < 0.769) (Supporting Information Table S1). Since the p-value calculation relies on the number of reads obtained for each gene, this procedure enabled us to take into account genes with low mRNA levels, such as many transcrip-tional regulators and co-regulators. Thus, genes considered as dif-ferentially expressed in this study include (a) genes that are statistically differentially expressed (p< .05), whatever the magni-tude of the fold-change is and (b) genes that display a fold-change greater than the indicated threshold (>1.3 or < 0.769), although with a reduced statistical conﬁdence (p< .25). Importantly, only differentially expressed genes (DEGs) displaying a simultanenous up- or downregulation in the three comparisons (differentiation vs. expansion, differentiation vs. day 0, and differentiation vs. undiffer-entiation within the same time point) were considered in our anal-yses (Fig. 2). The requirement for a gene to be selected in the three comparisons at the same time point enabled us to select the changes due to the cell differentiation per se and not the expres-sion changes due to the prolonged in vitro culture without serum (comparisons vs. undifferentiated cells) or change in conﬂuency (comparisons vs. expansion condition) or basal medium (compari-sons vs. expansion or vs. day 0).

Pathway and Gene Ontology Analyses

Data were analyzed using the QIAGEN’s Ingenuity Pathway Analysis suite (IPA, QIAGEN, www.qiagen.com/ingenuity). Gene ontology (GO)-TERMS annotations were analyzed using DAVID Bioinformatics Resources [11].

Identification of Predicted Co-Regulators of

Mitochondrial Biogenesis and Stem Cell Commitment

to Differentiation

Biological process-, cell component-, and molecular function-associated GO terms were retrieved for DEGs at day 5 of dif-ferentiation. DEGs were then sorted based on their GO anno-tations: one set was established with genes containing annotations related to mitochondria (“Mitochondria dataset,” Supporting Information Table S2), another one with genes containing annotations related to (mesenchymal) stem cells or differentiation (“Stem-diff dataset,” Supporting Information Table S3). The “upstream analysis” function of IPA was used to identify the putative regulators of these two lists of DEGs (Supporting Information Tables S4, S5). This function, which relies on a database containing the expected effects between transcriptional regulators and their target genes, retrieves the transcriptional regulators that are putatively involved in the differential expression of a list of genes. It also predicts whether these regulators are likely activated or inhibited, based on the expression value of their targets.

Real-Time Reverse Transcriptase-qPCR Analysis

RNA extraction, cDNA synthesis, and qPCR were performed as described previously [12] using the following primers: a1AT (a-1-antitrypsin) F: GGGTCAACTGGGCATCACTAA R: CCCTTTCT CGTCGATGGTCA; CPS1 (carbamoyl phosphate synthetase 1) F:

TCCTGATAGGCATCCAGCAATC R: AGGGACATTGTTGGCGTTGA;

PGC-1a (peroxisome proliferator-activated receptor gamma, coactivator 1 alpha) F: TCCTTTCTCTCGCCCAACACGATCT R: GC ATCCGACAGGACAAACAGTGGA; glutamine synthetase (GLUL) F:

CGCCGAGATGAGAAAGTTGT R: AAGCATGGCAGAGTGGGTAA;

DNA polymerase gamma (POLG) F: TGGCCTTGCAGATCACCAA

CCT R: TCCTTCCTGAGGCACCGGTCAA; POLG2 (DNA Polymerase Gamma 2, Accessory Subunit) F: TCGACTCCAGTGGTGGAGAAA

R: TTCCTTTCCGGCCTTCTTCAT; POLRMT (RNA Polymerase, Mitochondrial) F: GTGTACGGGGTCACGCGCTA R: CCGAGAACA TCTCCTGTAGACTCTT; PPIE (Peptidyl-prolyl cis-trans isomerase E) F: CCGCTCTTGACCCTGCATAT R: TCCAAGCAGACCCTGAGGAA; TBX3 (T-box transcription factor 3) F: GTCGGGAAGGCGAATG TTTC, R: ACGACAGTCATCAGCAGCTA; TCF7L2: F: CCTGGCACCGT AGGACAAATC R: GAGGGAACCTGGACATGGAAGC; TDO2 (tryp-tophan 2,3-dioxygenase) F: GAGGAACAGGTGGCTGAATTT R:GC TCCCTGAAGTGCTCTGTA. All results were normalized to the mRNA abundance of the housekeeping gene peptidyl-prolyl cis-trans isomerase E (PPIE).

Determination of the Relative Abundance of

Mitochondrial DNA

The abundance of mitochondrial DNA (mtDNA) was quantiﬁed as described previously [12].

Western Blot Analysis. Western blotting analyses were per-formed as described previously [12], using the following primary antibodies (Target, Manufacturer, Reference, Dilution factor): ATP synthasea, Life Technologies, A21350, 1:1,000; ATP synthase b, Life Technologies, A21351, 1:2,000; COX I (Cytochrome c oxidase, subunit 1), Abcam, Cambridge, UK, http://www.abcam.com, ab14705, 1:2000; COX II (Cytochrome c oxidase, subunit 2), Abcam, ab110258, 1:2000; COX IV (Cytochrome c oxidase, subunit 3), Life Technologies, A21348, 1:500; Histone 3, Cell Signaling Technology, Danvers, MA, https://www.cellsignal.com, #4499, 1:5,000; PGC-1a (peroxisome proliferator-activated receptor gamma, coactivator 1 alpha), Santa Cruz Biotechnology, Dallas, TX, https://www.scbt. com, SC-13067, 1:500; TCF7L2, Cell Signaling Technology, #2569, 1:1,000; ubiquinol-cytochrome c reductase, complex III subunit VII (UQCRQ), Proteintech, 14975–1-AP, 1:2,000.

Respiration Assays

The extracellular flux analyzer XF96 and the XF Cell Mito Stress Test Kit (Seahorse Biosience, North Billerica, MA) were used to perform metabolic profiling and assess mitochondrial function [13]. The day before the assay, equal number of cells was seeded in wells of a XF cell culture plate. A XF cell mito stress test (Sea-horse Bioscience, part of Agilent, North Billerica, MA) was per-formed according to the manufacturer’s instructions and using the following chemical concentrations: 1 lM oligomycin, 0.5 lM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), 1 lM antimycin A, and 1 lM rotenone. Once oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) measurements were performed, cells were lysed in 0.5 N NaOH, and OCR and ECAR measurements were normalized to the protein content of the well. Basal respiration was calculated by substracting the non-mitochondrial oxygen consumption [OCR following the addition of both antimycin A and rotenone (mitochondrial complex III and I inhibitors, respectively)] to the basal OCR. ATP coupling efficiency was calculated by substract-ing the non-mitochondrial oxygen consumption from the OCR following oligomycin addition (an ATP synthase inhibitor) and is expressed as a percentage of the basal respiration. Spare respi-ratory capacity was calculated by substracting the non-mitochondrial oxygen consumption from the OCR following FCCP (an uncoupling protonophore) addition and is expressed

(5)

as a percentage of the basal respiration. For each condition and each donor, 4–12 technical replicates were performed.

Immunofluorescence and Quantification of the

Mitochondrial Content per Cell

Cells were seeded on coverslips, ﬁxed with 4% paraformaldehyde for 20 minutes, permeabilized for 5 minutes with phosphate-buffered saline (PBS)-1% Triton X-100, and then incubated for 2 hours with an anti-TOM20 antibody (Santa Cruz Biotechnology, SC-11415, 1:200) (for mitochondria staining) and an anti-b-catenin antibody (BD Transduction Laboratories, part of BD Bioscience, San Jose, CA, http://www.bdbiosciences.com, BD610153, 1:100) (for cytoplasm staining) diluted in PBS-1% bovine serum albumin (BSA). Cells were then incubated for 1 hour with Alexa-labeled secondary antibodies (Molecular Probes, Eugene, OR, now part of Thermo Fischer, https://www.thermoﬁsher.com/us/en/home/brands/ molecular-probes/key-molecular-probes-products.html) (1:1,000) and observed by confocal microscopy (Leica Microsystems, Buffalo Grove, IL, Dako (now Agilent Technologies, Santa Clara, CA, http:// www.agilent.com/)). The area occupied by the mitochondrial stain-ing per cell was calculated usstain-ing the ImageJ software.

Immunohistochemistry

Five micrometer thick human liver sections at different stages were deparaffinized and rehydrated in graded alcohol series. Endogenous peroxidase activity was blocked by incubation for 30 minutes in a 1% hydrogen peroxide methanol solution. Antigen retrieval was performed by incubating the sections in citric acid monohydrate solution (pH 9.0) at 988C for 35 minutes. Non-specific immunostaining was prevented by 1 hour incubation in PBS buffer containing either 5% (immunochemistry) or 10% (immunofluorescence) BSA at room temperature. Slices were incubated for 1 hour with polyclonal antihuman TCF7L2 (Abcam; 1:500), in 0.5% BSA at 378C. After washing with PBS-Tween-20 0.5% solution, immunostaining detection was visualized by Envi-sion Dako anti-rabbit (Dako), using diaminobenzidine (Sigma) as chromogenic substrate. The nuclei were counterstained using Mayer’s hematoxylin for 10 minutes and mounted using Neo-Entellan (Merck, Darmstadt, Germany, http://www.merck.com) for microscopy analysis. The same protocol was used for immuno-fluorescence staining with TCF7L2 (Abcam; 1:500) and anti-TOM20 (Santa Cruz Biotechnology, SC-11415, 1:200) except that these primary antibodies were revealed by a 1 -hour incubation with the corresponding Alexa-labeled secondary antibodies (Molecular Probes) (1:1,000) before staining nuclei for 5 minutes with 40,6-diamidino-2-phenylindole. Slides were mounted in fluo-romount medium (Sigma).

Statistical Analyses

Depending on the data to compare, two-tailed paired t tests (Figs. 1D, 5G–5J) or two-way paired analysis of variance tests (Figs. 1A, 1B, 1C, 3, 4, 5C) were performed, assuming normal distribution of data. In Figure 5G, data were log(2) transformed to ensure homogeneity of variance. To identify the signiﬁcantly different groups, pairwise multiple comparisons using the Student–Newman–Keuls method were performed (* or #, p< .05; ** or ##, p< .01; *** or ###, p < .001) (when used on a same chart: *, comparisons of Diff vs. day 0; #, comparisons between Diff and Undiff at a particular time point). All statistical analyses were carried out using the SigmaPlot 12.5 software (Systat Soft-ware, San Jose, CA, https://systatsoftware.com).

R

ESULTS

The Mitochondrial Biogenesis and Metabolic Shift

Toward OXPHOS Overlap the Decreased Expression of

Genes Involved in Pluripotency Pathways During the

Commitment of BM-MSCs to the Hepatocytic Lineage

We previously demonstrated a strong mitochondrial biogenesis, characterized by an increase in the abundance of mitochondria, mtDNA and several OXPHOS subunits, in a model of hepatocytic differentiation of human BM-MSCs [12]. To get a deeper insight into the mechanisms regulating this mitochondrial biogenesis, we monitored these parameters over time. We evaluated the abundance of mtDNA, of several OXPHOS subunits (the UQCRQ), the subunits I, II, and IV of the cytochrome C oxidase (COX I, COX II, and COX IV) and the subunitsa and b of the ATP synthase and the respiration rate throughout the hepatocytic induction step (from day 0 to day 12 of the differentiation pro-cess, lasting 22 days in total) (Fig. 1). A slight but significant induction of mtDNA was detected since the 5th day of differen-tiation (Fig. 1A). Similarly, several OXPHOS subunits were signifi-cantly more abundant after 5 days of differentiation (Fig. 1B, 1C). This resulted, at the functional level, in a significantly increased basal respiration, coupling efficiency (fraction of the basal respiration that is used for ATP production) and spare respiratory capacity (cell’s ability to increase substrates and electron flux across the OXPHOS chain to respond to an increased energy demand) of the OXPHOS system, as soon as 5 days of differentiation (Fig. 1D). In addition, the basal ECAR, an undirect measurement of glycolysis, was lower in differentiated BM-MSCs (Fig. 1D). These data support that the mitochondrial biogenesis and metabolic shift toward OXPHOS occurred within the first 5 days of hepatocytic differentiation.

To unveil the mechanisms regulating mitochondrial biogene-sis during the differentiation of BM-MSCs into hepatocyte-like cells, we performed transcriptomics analyses at early time points of the differentiation process. Strikingly, analysis of the 10 most enriched pathways among genes upregulated after 5 days of differentiation revealed an enrichment of the “OXPHOS pathway” (Fig. 1E, right panel). Indeed, 30 of 104 (29%) of the OXPHOS subunits referenced in this pathway were upregulated after 5 days of differentiation. These data support the notion that the observed increased OXPHOS activity (Fig. 1D) is at least partly explained by a broad induction of OXPHOS-related genes at the transcript level. We also observed an induction of the “Mitochondrial dysfunction pathway,” a pathway that includes mitochondrial proteins frequently mutated or dysregulated in mitochondrial diseases, including OXPHOS subunits, metabolic enzymes (e.g., TCA cycle) and ROS detoxifying enzymes. This observation supports the notion that many mitochondrial genes—besides OXPHOS—are upregulated during differentia-tion. The upregulation of ROS detoxifying enzymes has also been reported during the osteogenic differentiation of MSC and would protect the differentiating cells from the increased gener-ation of ROS accompanying the switch toward OXPHOS metabo-lism [14]. In agreement with the progressive acquisition of a hepatic fate, an induction of metabolic pathways related to cho-lesterol synthesis, a typical hepatic function, was also observed. On the other hand, pathways involved in the regulation of stem-ness and pluripotency, including the “human ESC pluripotency,” “Wnt/b-catenin signaling,” and “transforming growth factor

(6)

Figure 1. Mitochondrial biogenesis and metabolic shift toward oxidative phosphorylation (OXPHOS) occur early during the hepatocytic dif-ferentiation. (A): Mitochondrial DNA abundance. (B): Abundance of several OXPHOS subunits analyzed by Western blot. (C): Quantification of the signal intensity. (D): Metabolic profiling and assessment of the mitochondrial function of differentiated and undifferentiated cells at day 5. (E): Top 10 enriched pathways among upregulated and downregulated genes after 5 days of differentiation, ranked according to increasing p values of the enrichment. (A, C): Data are presented as the mean6 SEM, relative to cells in expansion. (D): Data are presented as the mean6 SEM. (A–C) n 5 3, (D) n 5 4 independent experiments. * or #, p < .05; ** or ##, p < .01; *** or ###, p < .001; *, comparisons versus day 0; #, comparisons between Diff and Undiff. Abbreviations: ECAR, extracellular acidification rate; EIF2, eukaryotic translation initia-tion factor 2; FCCP, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone; ILK, integrin linked kinase; mTOR, mammalian target of rapamy-cin; OCR, oxygen consumption rate; TGF, transforming growth factor.

(7)

beta (TGF-b) signaling” pathways, were strongly enriched among the genes downregulated after 5 days of differentiation (Fig. 1E, left panel). Thus, the induction of mitochondrial biogen-esis overlaps the downregulation of genes belonging to path-ways involved in the control of stem cell pluripotency, during the early hepatocytic differentiation of BM-MSCs. Supporting the loss of the mesenchymal phenotype and promotion of the hepatic differentiation, a downregulation of genes involved in the “hepatic ﬁbrosis pathway “(that includes the TGF-b pathway - an inhibitor of hepatocyte proliferation and differentiation [15]- and downstream of it, expression of collagens) was observed. Similarly, due to the important change in morphology (cell elongation) observed during the ﬁrst step of the differentia-tion process, the “epithelial adherens juncdifferentia-tion pathway” (involv-ing members of the TGF-b pathway, catenins, tubulins, and actins) was also statistically downregulated.

Identification of Predicted Co-Regulators of

Mitochondrial Biogenesis and Stem Cell Commitment

to Differentiation

A lack of understanding currently resides in the mechanisms connecting, on the one hand, the increase in mitochondrial biogenesis and function and, on the other hand, the loss of pluripotency and differentiation of stem cells. We hypothe-sized that these two concomitant processes might be con-trolled by common regulators. We thus investigated the putative upstream regulators of DEGs (Supporting Information Table S1) related to the mitochondria (Supporting Information Table S2) and to stem cell pluripotency or differentiation (Sup-porting Information Table S3), seeking to identify putative common regulators of the two processes (Supporting Informa-tion Table S4, S5, respectively) (the strategy developed to identify those common regulators is detailed in the Material and Method section and illustrated in Fig. 2).

Using IPA, 40 predicted co-regulators were identified, of which 9 displayed after 3 and/or 5 days of differentiation a dif-ferential expression consistent with their predicted activation status (Table 1). Among them, HIF-1a is the candidate predicted to be the most strongly repressed during differentiation, while PGC-1a, predicted to be activated during differentiation, is the candidate displaying the highest differential expression after 5 days of differentiation (Table 1). Noteworthy, the efficiency of our bioinformatics approach in identifying co-regulators of mitochondrial biogenesis and cell commitment to differentia-tion is supported by the identificadifferentia-tion of HIF-1a as one of those putative co-regulators. Indeed, in ESCs, HIF-1a positively regu-lates pluripotency while inhibiting mitochondrial biogenesis and function, through its binding to the promoter of pluripotency genes and PGC-1a, resulting in transcriptional activation and inhibition, respectively [4]. Since HIF-1a is not expressed at the protein level in adult BM-MSCs (unpublished personal data and [16]), we decided to further focus on 1a. Importantly, PGC-1a is reported to interact with multiple transcriptional regula-tors also identified as putative co-regularegula-tors of mitochondrial biogenesis and stem cell differentiation, including PPAR-a [17], PPAR-g [18], the hepatocyte nuclear factor-4a (HNF-4a), and the glucocorticoid receptor (GR or NR3C1) [19] (Table 1). These observations suggest that PGC-1a might exert a broad effect on the co-regulation of mitochondrial biogenesis and cell differen-tiation. We thus investigated whether PGC-1a induction was

required for the hepatocytic differentiation of BM-MSCs and the associated mitochondrial biogenesis.

Repressing PGC-1a Induction Partly Prevents

Mitochondrial Biogenesis and Hampers BM-MSC

Hepatocytic Differentiation

PGC-1a mRNA is strongly induced during hepatocytic differenti-ation (Fig. 3A), resulting in a modest but signiﬁcant increase in the protein level (day 5 Diff vs. day 0: p5 .032; day 12 Diff vs. day 0: p5 .002; day 22 Diff vs. day 0: p 5 .003) (Fig. 3D, 3E). To test whether the1.5-fold increase in PGC-1a abundance was responsible for the increased mitochondrial biogenesis and function, we used shRNA against PGC-1a. We repressed the induction of PGC-1a mRNA by 54%, 52% and 64% after 5, 12 and 22 days of differentiation, respectively (Fig. 3A). At the

pro-tein level, we inhibited by 93%, 53% and 49% the

differentiation-induced increases in PGC-1a expression observed at these respective time points (Fig. 3E). The fact that we only partially repressed PGC-1a induction, without repres-sing its abundance to lower levels than what is observed in expanding stem cells (i.e., not lower than the fold change tres-hold set to 1), allowed us to assess the requirement of PGC-1a induction, speciﬁcally (rather than basal expression), for the mitochondrial biogenesis and commitment to the hepatocytic lineage.

The repression of PGC-1a led to a modest repression of mtDNA relative abundance, mainly in undifferentiated cells and cells differentiated for 5 days (Fig. 3B). To determine whether other regulators were involved in the regulation of mtDNA abundance, we analyzed the mRNA abundance of several major regulators of mtDNA replication, including POLG, which enco-des the catalytic core of the DNA polymerase g involved in mtDNA replication; TFAM, a mitochondrial transcription factor whose expression has been correlated with mtDNA abundance [20] and of the mitochondrial RNA polymerase POLRMT, which is involved in mtDNA replication by synthetizing an RNA primer required for the initiation of mtDNA replication [21]. Modest changes were observed in the expression of those regulators during differentiation (Supporting Information Fig. S1). More-over these changes were not consistent with the mtDNA induc-tion proﬁle observed during differentiainduc-tion. More interestingly, we found that the mRNA encoding the DNA polymerase g accessory subunit POLG2, was increased 1.5- to twofolds during the hepatocytic differentiation. Incidentally, we found that the repression of PGC-1a led to a modest but signiﬁcant repression of POLG2 induction (Fig. 3B). Altogether, these data suggest that PGC-1a regulates POLG2 expression, regulating itself mtDNA replication. However, the partial inhibition of PGC-1a induction - and hence, of POLG2 induction, only leads to minor differences in mtDNA abundance.

In contrast, the repression of PGC-1a drastically prevented the increase in mitochondrial content per cell observed by the end of the hepatocytic induction step (Fig. 3C). The strong impact of PGC-1a on mitochondrial abundance might be related to its ability to co-activate several families of nuclear transcription factors involved in mitochondrial biogenesis [5]. Similarly, PGC-1a repression was associated with a signiﬁcant decrease in the protein abundance of the subunitsa and b of the ATP synthase and of the subunit UQCRQ of the complex III (Fig. 3D, 3E). Altogether, the partial repression of PGC-1a

(8)

Figure 2. Strategy developed to identify common transcriptional regulators of gene expression changes related to stem cell differentiation and mitochondria. Transcriptomics analyses were performed at early time-points of the differentiation process and differentially expressed genes (DEGs) were identified (see Material and Method section for further details). Gene ontology (GO) terms were retrieved for DEGs at day 5 of differentiation. DEGs were then sorted based on their GO annotations: one set was established with genes containing annotations related to mitochondria (“Mitochondria dataset”), another one with genes containing annotations related to (mesenchymal) stem cells or differentiation (“Stem-diff dataset”). The “upstream analysis” function of IPA was used to identify the putative regulators of these two lists of DEGs. This function, which relies on a database containing the expected effects between transcriptional regulators and their target genes, retrieves the transcriptional regulators that are putatively involved in the differential expression of a list of genes. It also predicts whether these regulators are likely activated or inhibited, based on the expression value of their targets. The common regulators of the Stem-diff dataset and mitochondria dataset were subsequently selected, and their expression profile at day 3 and/or 5 of differentiation was retrieved (see Table 1). Abbreviations: DEG, differentially expressed gene; EGF, epidermal growth factor; FGF, fibroblast growth factor; GO, gene ontol-ogy; HGF, hepatocyte growth factor; ITS, insulin–transferrin–selenium; MACE, massive analysis of cDNA ends; OSM, oncostatin M.

(9)

1. Identiﬁcation of common pr edict ed regulator s o f the diff ere ntially e xpr essed genes with GO relat ed to mitochondria and st e m cell diff er entiation a nd comparison of their pr edict ed activ ation st atus e xpr ession at the mRNA level T ranscriptional _regulator (TR) Pr edict ed activ ation status of the TR based on g ene e xpr ession of the mitochondria dataset Pr edict ed activ ation status of the TR based on g ene e xpr ession of the St em-diff dataset Diff er ential e xpr ession of the TR aft er 3 d a y s of diff ere ntiation Diff ere ntial e xpres sion of the TR after 5 d a y s of differ entiation Activ ation z-scor e p v alue of overlap Ta rg ets Activ ation z scor e p v alue of overlap Ta rg ets Da y 3 dataset Log2FC (da y 3 diff vs. e xp) p v alue (da y 3 diff vs. e xp) Day 5 dataset Log2FC (da y 5 diff vs. e xp) (da vs. er entia l e xpr essi on is tent with the edict ed acti v atio n at us P GC-1 a 2.21 1.25E -13 19 1.43 2.39E -04 10 / 1.15 2.2 2E -0 1 Upr egulat ed 3.25 4.36E P P AR-g 2.60 4.93E -07 19 1.41 6.92E -21 40 Upr egul at ed 1.68 1.2 6E -1 5 Upr egulat ed 1.79 6.69E SR EBF1 1.12 2.29E -03 8 1.57 1.74E -06 14 / 1.22 1.9 6E -3 8 Upr egulat ed 1.66 1.38E IFI 16 1.66 1.02E -04 6 1.28 4.88E -06 8 Upr egul at ed 0.47 1.9 7E -1 4 Upr egulat ed 0.97 2.17E ST A T 6 1.46 3.94E -02 6 1.13 1.47E -07 16 Upr egul at ed 0.50 8.7 9E -1 3 Upr egulat ed 0.38 3.45E NU PR1 2 1.47 3.18E -02 12 2 1.07 3.15E -04 20 Downr egul at ed 2 1.01 2.7 3E -1 79 Dow nr egulat ed 2 0.46 1.78E HI F-1 a2 1.43 1.56E -08 19 2 2.39 1.54E -26 42 Downr egul at ed 2 1.01 4.8 0E -1 75 Dow nr egulat ed 2 1.05 7.88E E GR2 2 0.34 7.03E -03 6 2 0.62 5.29E -10 16 Downr egul at ed 2 0.98 2.4 4E -0 5 Dow nr egulat ed 2 2.11 9.59E ST A T 4 2 0.66 2.26E -02 7 2 0.59 6.86E -05 13 Downr egul at ed 2 2.09 1.6 3E -1 1 Dow nr egulat ed 2 2.19 1.45E essi on chang e is tent with the edict ed acti v atio n at us but not at is tically sign iﬁcant all com pariso ns MY CN 1.32 1.24E -03 10 0.93 4.28E -09 20 / 0.83 7.6 8E -0 1 / 1.12 6.45E P P AR-a 2.07 2.11E -11 26 1.85 1.92E -08 25 / 0.68 1.5 8E -0 4 / 0.26 1.29E HN F4-a 3.13 2.24E -10 61 1.70 3.27E -02 44 / 0.83 7.6 8E -0 1 / 0.12 9.68E VH L 1.89 3.75E -06 10 1.70 1.45E -09 15 / 0.25 8.3 9E -0 2 / 0.07 5.97E SIR T1 2 0.24 3.60E -04 7 2 0.98 1.01E -06 11 / 2 0.12 5.3 3E -0 1 / 2 0.19 2.63E MY C 2 0.37 1.17E -14 43 2 0.24 1.34E -22 47 / 0.15 8.2 9E -0 2 / 2 0.35 3.53E ES R2 2 1.20 1.62E -02 7 2 2.21 3.26E -12 22 / 2 0.17 9.4 4E -0 1 / 2 0.88 7.16E ect ed mRNA C/ EBP -a 0.99 2.30E -02 10 1.31 1.35E -10 18 / N A N A / NA G A TA 1 2.00 1.65E -02 6 0.64 9.99E -11 19 / N A N A / NA G A TA 3 1.97 3.93E -02 4 0.27 2.73E -07 11 / N A N A / NA G A TA 4 1.18 2.90E -02 4 0.08 5.66E -06 10 / N A N A / NA NR 1H4 2 1.89 5.76E -02 4 2 1.18 2.61E -06 11 / N A N A / NA PD X 1 2 0.38 4.56E -04 9 2 0.58 2.77E -14 32 / N A N A / NA essi on chang e cons is te nt wi th pr edict ed at ion st atus ES R1 2 0.11 2.62E -02 10 2 0.82 5.89E -19 38 / 0.37 5.6 8E -0 1 / 1.12 3.07E P GC-1 b2 1.11 3.36E -03 4 2 0.57 1.14E -03 5 / 0.10 9.2 5E -0 1 / 0.38 6.63E HTT 2 1.45 1.07E -09 32 2 0.40 3.21E -15 46 Upr egul at ed 0.33 2.6 6E -0 2 Upr egulat ed 0.28 4.10E TP5 3 0.97 4.76E -15 54 0.34 1.13E -34 47 Downr egul at ed 2 1.10 3.5 3E -2 1 / 2 0.12 1.79E FO X O 3 2.24 1.02E -05 10 0.02 5.39E -08 14 Downr egul at ed 2 0.78 8.1 7E -0 9 / 2 0.17 1.11E HS F1 0.31 7.36E -06 11 1.41 4.34E -07 14 / 2 0.29 7.4 8E -0 2 / 2 0.23 1.15E ME F2C 1.20 1.77E -03 5 0.39 8.56E -04 6 / 2 0.25 3.7 6E -0 1 / 2 0.28 2.62E EP AS1 1.97 1.38E -02 6 0.54 5.67E -17 24 Downr egul at ed 2 1.49 9.9 3E -2 26 / 2 0.40 7.48E NF E2L2 2.43 4.28E -04 14 0.25 9.47E -08 24 Downr egul at ed 2 0.47 1.3 1E -1 1 Dow nr egulat ed 2 0.41 2.32E ET S2 1.22 6.72E -03 4 0.23 4.88E -06 8 Downr egul at ed 2 0.86 1.4 9E -0 2 / 2 0.46 1.24E NR 3C1 0.32 6.51E -04 19 1.16 4.18E -17 46 Downr egul at ed 2 0.30 1.0 4E -0 6 Dow nr egulat ed 2 0.65 2.92E RE L 1.22 1.75E -02 5 1.64 6.81E -08 13 / 0.05 9.3 4E -0 1 / 2 0.66 3.27E RB1 1.15 1.93E -02 8 0.48 2.88E -14 27 / 2 0.30 1.8 1E -0 2 Dow nr egulat ed 2 0.79 1.02E MU C1 2.22 2.93E -05 5 0.94 1.41E -08 8 Downr egul at ed 2 1.90 8.2 1E -1 1 Dow nr egulat ed 2 0.84 1.42E FO XA1 0.30 1.27E -02 5 2.13 2.73E -09 14 / 0.83 6.7 7E -0 1 / 2 0.88 7.16E WT 1 0.15 3.42E -02 5 0.89 1.59E -13 20 / 0.83 6.7 7E -0 1 / 2 0.88 7.16E G FI1 1.00 2.69E -02 4 1.80 2.22E -08 12 2 1.75 3.3 2E -0 1 2 1.47 3.44E NR 4A3 1.11 5.41E -05 6 2.38 2.24E -05 7 / 2 4.38 6.6 3E -0 8 / 2 2.77 5.45E

(10)

chondrial DNA abundance and DNA polymerase gamma-2 mRNA abundance. (C): Immunofluorescence of the mitochondrial network, visualized by a TOM20 immunostaining (red). Cells are counterstained using ab-catenin immunostaining (cyan blue). The mitochondrial content per cell was determined by calculating the area, per cell, occupied by the mitochondrial network. Ten cells were analyzed per condition and per experiment, n5 3 independent experiments. Scale bar 5 50lm. (D): Abundance of several oxidative phosphorylation subunits analyzed by Western blot. (E): Quantification of the Western blot signal intensity. (F): Morphology of cells at the end of the dif-ferentiation process (day 22), observed by phase-contrast microscopy. Scale bar5 250 lm. (G): mRNA abundance of differentiation markers. (A, B, E, G): Data are presented as the mean6 SEM, relative to untransduced cells in expansion. (C): Data are presented as the mean6 SEM. (A, B, D, E, F, G): n 5 5 independent experiments. For clearness, only significant differences between transduction condi-tions (shNT vs. shPGC-1a) were illustrated (*, p < .05; **, p < .01; ***, p < .001). Abbreviacondi-tions: CPS1, carbamoyl phosphate synthetase 1; GLUL, glutamine synthetase; PGC-1a, peroxisome proliferator-activated receptor gamma coactivator-1 alpha; shNT, non-target control

(11)

induction partly impaired the mitochondrial biogenesis observed during the hepatocytic differentiation of BM-MSCs.

We next asked whether limiting the induction of PGC-1a would also disturb the hepatocytic differentiation or not. Strikingly, the acquisition of the typical hepatocyte-like polyg-onal shape, normally observed by the end of the differentia-tion process, was strongly hampered in cells transduced with shRNA against PGC-1a (Fig. 3F). At the mRNA level, the induc-tion of TBX3, a transcripinduc-tion factor expressed in hepatoblasts [22] and induced during the early commitment to hepatocytic differentiation, was completely abrogated in shRNA against PGC-1a (shPGC-1a)-transduced cells (Fig. 3G). Moreover, a delayed and reduced induction of the hepatocytic marker a-1-antitrypsin (a1AT) was observed in shPGC-1a-transduced cells. Similarly, a reduced induction of the hepatic/periportal gene CPS1 and of the pericentral marker glutamine synthase (GLUL) was observed by the end of the differentiation process (Fig. 3G). Altogether, these results support that PGC-1a is required

for the mitochondrial biogenesis, for the morphological matu-ration into hepatocyte-like cells and for the acquisition of hepatic gene expression of both periportal and perivenous regions.

OXPHOS Activity Is Required for the Differentiation

into Hepatocyte-like Cells

PGC-1a co-activates several transcription factors involved in liver development and hepatic gene expression, including CCAAT/enhancer binding protein b [23], HNF-4a [19], the forkhead box protein O1 [24], and the glucocorticoid receptor [25]. Therefore, PGC-1a might contribute to hepatocytic dif-ferentiation directly, by transactivating hepatic transcription factors, and/or undirectly, by inducing mitochondrial biogene-sis and function. To test the latter possibility, we differentiated BM-MSCs in the presence of inhibitors of the mitochondrial complex III (myxothiazol and antimycin A) and V (oligomycin). Inhibiting OXPHOS activity prevented the acquisition of the

Figure 4. Inhibiting oxidative phosphorylation hampers hepatocytic differentiation. (A): Morphology of cells at the end of the differen-tiation process (day 22), observed by phase-contrast microscopy. Scale bar5 250lm. (B): mRNA abundance of differentiation markers and peroxisome proliferator-activated receptor gamma coactivator-1 alpha. Data are presented as mean6 SEM (n5 4 independent experiments) relative to untreated cells in expansion. For clearness, only signiﬁcant differences between molecules and their vehicle were illustrated (*, p< .05; **, p < .01; ***, p < .001). Abbreviations: CPS1, carbamoyl phosphate synthetase 1; DMSO, dimethyl sulfox-ide; GLUL, glutamine synthetase; PGC-1a, peroxisome proliferator-activated receptor gamma coactivator-1 alpha.

(12)

hepatocyte-like polygonal shape (Fig. 4A). Importantly, the induction of the hepatocytic markers a1AT and tryptophan 2,3-dioxygenase (TDO2) is strongly reduced upon OXPHOS inhibition (Fig. 4B). Myxothiazol treatment also inhibited the induction of the periportal gene CPS1 and of the pericentral marker GLUL (Fig. 4B). Importantly, the inhibitors did not induce a strong ATP shortage (Supporting Information Fig. S2) nor enhanced cell apoptosis (M Caruso, unpublished observa-tions), supporting the requirement for mitochondrial activity, speciﬁcally, for hepatocytic differentiation. Since the induction of PGC-1a mRNA is enhanced in the presence of the different OXPHOS inhibitors (Fig. 4B)—likely as a compensatory mecha-nism for the decreased OXPHOS activity—these data support that PGC-1a contributes to hepatocytic differentiation at least partly through its ability to induce mitochondrial biogenesis and function. Nonetheless, we did not observe any change in TBX3 induction in the presence of the OXPHOS inhibitors (unpublished observations), unlike upon PGC-1a repression (Fig. 3G). These observations suggest that PGC-1a might in addition stimulate hepatocytic differentiation independently of OXPHOS.

The TCF7L2-PGC-1a Axis Links Mitochondrial

Biogenesis and Metabolic Shift to the Early

Commitment to Hepatocytic Differentiation

We next focused deeper on the upstream regulators of the interplay between stem cell differentiation and mitochondrial biogenesis. We hypothesized that the expression of PGC-1a itself might be controlled by regulators of stem cell differenti-ation. Using IPA, we retrieved the transcriptional regulators of PGC-1a expression. We selected those that were differentially expressed after 3 or 5 days of differentiation and which expression changes were consistent with the induction of PGC-1a, based on litterature data (Fig. 5A; Supporting Infor-mation Table S6). TCF7L2 appeared as a strong candidate as it is a downstream effector of the Wnt/b-catenin signaling path-way [26] predicted to negatively affect PGC-1a expression according to IPA. Interestingly, systems biology network analy-ses revealed that TCF7L2 is functionally associated with the stemness regulator SOX2 in human ESCs [27], emphasizing its possible function as a molecular node connecting stem cell loss of pluripotency to the initiation of mitochondrial biogene-sis. In addition, the repression of Wnt/b-catenin signaling is required for early liver speciﬁcation [28]. Accordingly, we observed a strong repression of TCF7L2 expression, particu-larly at early time points of the hepatocytic differentiation (Fig. 5B, 5C), as well as a global downregulation of genes involved in the Wnt/b-catenin signaling pathway, after 5 days of differentiation (Fig. 1E). Besides, we evidenced a progres-sive downregulation of TCF7L2 expression during human liver development (Fig. 5D, 5E). Thus, we hypothesized that the downregulation of the Wnt/b-catenin signaling and of TCF7L2 expression during hepatic differentiation and maturation might trigger PGC-1a induction, mitochondrial biogenesis and metabolic shift. Supporting this hypothesis, we found that in neonatal livers, hepatocytes with a high TCF7L2 staining in the nuclei had a very limited mitochondrial network, while those displaying the largest mitochondrial networks had no detectable expression of TCF7L2 (Fig. 5F).

To test this hypothesis, we repressed TCF7L2 in expanding BM-MSCs and assessed whether TCF7L2 repression would be

sufﬁcient to induce mitochondrial biogenesis, independently of the stimulation with differentiation cocktails. Using lentiviral-mediated transduction of shRNA against TCF7L2, we repressed TCF7L2 mRNA and protein abundance by 56% and 49%, respec-tively (Fig. 5G–5I). In agreement with our hypothesis, TCF7L2 repression resulted in an increased abundance of PGC-1a tran-script (Fig. 5G) and in a 17% increase in PGC-1a protein abun-dance (Fig. 5H, 5I). Furthermore, the repression of TCF7L2 resulted in a signiﬁcant increase in the abundance of several OXPHOS subunits, including UQCRQ, COX I, COX II, COX IV, and the subunit b of the ATP synthase (Fig. 5H, 5I) and in an increased basal respiration rate (Fig. 5J). Altogether, TCF7L2 repression increases PGC-1a expression and OXPHOS activity in BM-MSCs.

D

ISCUSSION

In this study, we postulated that mitochondrial biogenesis and stem cell differentiation might be controlled by common tran-scriptional regulators. We found that PGC-1a was necessary for the induction of mitochondrial biogenesis and also for the acqui-sition of hepatic gene expression during hepatic differentiation.

As previously mentioned, PGC-1a co-activates several hepatic transcription factors, such as CCAAT/enhancer binding proteinb [23], HNF-4a [19], the forkhead box protein O1 [24], and the glucocorticoid receptor [25], in addition to its ability to co-activate mitochondrial transcriptional regulators such as the estrogen-related receptor a (ERRa) or nuclear respiratory factor-2 [5]. While we evidenced that OXPHOS activity is required for the hepatocytic differentiation of BM-MSCs, further studies are required to investigate the potential mitochondrial-independent role of PGC-1a in that process. PGC-1a has been demonstrated to regulate different loci through its interaction with mitochondrial or hepatic transcription factors [23]. There-fore, it would be interesting to investigate the loci and transcrip-tion factors bound by PGC-1a at different time points of the hepatocytic differentiation of BM-MSCs, to further elucidate its involvement in the co-regulation of hepatocytic differentiation and mitochondrial biogenesis. PGC-1a is also reported to regu-late chondrogenesis by co-activating the transcription factor SRY-related high mobility group-Box gene 9 (Sox9) in MSCs [30]. These data suggest that PGC-1a may co-regulate mitochondrial biogenesis and stem cell differentiation into several lineages, by co-activating different transcriptional regulators.

Although we restricted our studies to the transcriptional con-trol of PGC-1a, it is worth mentioning that post-translational modi-fications of PGC-1a affect its activity, including its ability to coactivate different transcription factors. For example, the phos-phorylation of PGC-1a downstream of the S6 kinase, activated upon feeding in the liver, reduces the ability of PGC-1a to coacti-vate HNF-4a, but not ERRa or PPARa [31]. This modification alters the ability of PGC-1a to control the expression of neoglucogenic but not mitochondrial genes [31]. Therefore, post-translational modifications of PGC-1a likely control its ability to regulate mito-chondrial biogenesis and stem cell differentiation toward different lineages. This remains a complex area of research, given the high number of post-translational modifications PGC-1a is subjected to [32].

Upstream of PGC-1a, we found that TCF7L2 negatively con-trols mitochondrial biogenesis and metabolic shift toward

(13)

tional regulators of PGC-1a (PPARGC1A) were retrieved using IPA. The regulators displaying a differential expression at day 3 and/or day 5 of differentiation (purple, downregulated; yellow, upregulated) consistent with the induction of PGC-1a are illustrated. (B): Abundance of TCF7L2 analyzed by Western blot. (C): Quantification of the Western blot signal intensity. Data are expressed relative to expanding cells and presented as mean6 SEM (n 5 6 independent experiments). (D): TCF7L2 expression was visualized by immunohistochemistry on sections from two independent fetal livers and adult livers. (E): TCF7L2 expression was visualized by immunofluorescence on sections from fetal, neonatal, and adult livers. Nuclei were stained with 40,6-diamidino-2-phenylindole (DAPI). (F): TCF7L2 and TOM20 were ana-lyzed by immunofluorescence on sections from neonatal livers. Nuclei were stained with DAPI. Arrows indicate cells with large mito-chondrial networks and no expression of TCF7L2. (G–J): Cells were transduced with non-target control shRNA (white bars) or specific shRNA against TCF7L2 (shTCF7L2) (grey bars). (G): mRNA abundance of TCF7L2 and PGC-1a. (H, I): Abundance of proteins analyzed by Western blot and quantification of the signal intensity. (J): Basal respiration determined by an extracellular flux analyzer. (G–J): Data are presented as mean6 SEM of n 5 7 independent experiments. * or #, p < .05; ** or ##, p < .01; *** or ###, p < .001. For (C), *, compar-isons versus day 0; #, comparcompar-isons between Diff and Undiff. Abbreviations: CV, central vein; OCR, oxygen consumption rate; PGC-1a,

(14)

OXPHOS. TCF7L2 is strongly repressed during the ﬁrst days of the hepatocytic differentiation process, suggesting that its repression would be particularly important to initiate the induc-tion of PGC-1a expression. PGC-1a is known to control its own expression in an auto-regulated manner [5], which may explain its sustained induction throughout the differentiation process, despite a weaker repression of TCF7L2 at later stages. Interest-ingly, previous data support a role for the downregulation of the Wnt/ß-catenin signaling pathway both in vivo [28], for the early stages of liver development and in vitro, for the hepatocytic dif-ferentiation of MSCs [33]. In umbilical cord blood-derived MSCs, the downregulation of FRIZZLED 8, a key component of the Wnt/b-catenin signaling pathway, was demonstrated to reduce TCF7L2 expression and favor the expression of several hepato-cytic markers, including albumin, C/EBPa, and CYP1A1/2 [33]. While it was previously unclear how the downregulation of Wnt/b-catenin and TCF7L2 might contribute to the differentia-tion of hepatocytes, our data support the hypothesis that the TCF7L2-mediated control of PGC-1a expression, mitochondrial biogenesis and activity, are at least partly involved. Importantly, the activation of canonical Wnt signaling was reported to inhibit PGC-1a expression and the differentiation of brown adipocytes [34], suggesting that the Wnt/b-catenin-PGC-1a axis might con-nect mitochondrial biogenesis with stem cell differentiation into different lineages.

C

ONCLUSION

Although mitochondria and metabolism are now considered as critical regulators of the pluripotency and differentiation of stem cells [1, 29], the molecular mechanisms regulating the crosstalk between these two processes remain elusive. In this study, we provided data supporting that TCF7L2 repression upon commit-ment to hepatocytic differentiation de-represses mitochondrial biogenesis and triggers the differentiation-associated metabolic shift toward OXPHOS, by inducing PGC-1a expression. In turn, PGC-1a and OXPHOS activity further contribute to the differentia-tion to the hepatocytic lineage, by regulating the expression of hepatocytic genes and supporting the morphological maturation associated with the differentiation process.

Importantly, better deciphering the interplay between mitochondria and stem cell differentiation should improve our

ability to manipulate stem cells in vitro, which is of great interest for the development and optimization of stem cell-based therapeutics. In addition, since mitochondria regulate the pluripotency and function of stem cells, which are com-promised upon aging and diseases, understanding this inter-play might also lead to the development of regenerative medicine and novel anti-aging therapies.

A

CKNOWLEDGMENTS

We thank Pr. Sonveaux group (FATH, IREC, Universite Catholi-que de Louvain, Belgium) for their help with respiration analy-ses, members of Dr N. Tifﬁn’s group and SANBI (SANBI, University of the Western Cape, South Africa) for their warm welcome and helpful advice with bioinformatics analyses, and E. Di Valentin (University of Lie` ge, Belgium) for his help with lentiviral vectors. This work was supported by fellowship from (Fonds National de la Recherche Scientiﬁque [FNRS], Belgium) (to A.W.) and (Fonds de la Recherche dans l’Industrie et l’Agri-culture [FRIA]) (to M.C.).

A

UTHOR

C

ONTRIBUTIONS

A.W.: conceived and designed the study, performed experi-ments and collected data, analyzed and interpreted the data, wrote the manuscript; M.C.: performed experiments and col-lected data, analyzed and interpreted the data; J.-B.D.E., A.F., C.D., J.E., and H.E.-K.: performed experiments and collected data; M.N., G.P., and E.S.: provided administrative support and study material; T.A. and N.T.: analyzed and interpreted the data, provided administrative support and study material; M.N.: conceived and designed the study, analyzed and inter-preted the data, provided administrative support and study material; P.R.: conceived and designed the study, analyzed and interpreted the data, wrote the manuscript, provided administrative support and study material.

D

ISCLOSURE OF

P

OTENTIAL

C

ONFLICTS OF

I

NTEREST

The authors indicated no potential conﬂicts of interest.

R

EFERENCES

1 Xu X, Duan S, Yi F et al. Mitochondrial regulation in pluripotent stem cells. Cell Metabol 2013;18:325–332.

2 Mandal S, Lindgren AG, Srivastava AS et al. Mitochondrial function controls prolif-eration and early differentiation potential of embryonic stem cells. STEM CELLS 2011;29: 486–495.

3 Varum S, Momcilovic O, Castro C et al. Enhancement of human embryonic stem cell pluripotency through inhibition of the mito-chondrial respiratory chain. Stem Cell Res 2009;3:142–156.

4 Mahato B, Home P, Rajendran G et al. Regulation of mitochondrial function and cel-lular energy metabolism by protein kinase C-lambda/tau: A novel mode of balancing pluri-potency. STEMCELLS2014;32:2880–2892.

5 Hock MB, Kralli A. Transcriptional control of mitochondrial biogenesis and function. Ann Rev Physiol 2009;71:177–203.

6 Scarpulla RC, Vega RB, Kelly DP. Tran-scriptional integration of mitochondrial bio-genesis. Trend Endocrinol Metabol 2012;23: 459–466.

7 Tormos KV, Anso E, Hamanaka RB et al. Mitochondrial complex III ROS regulate adi-pocyte differentiation. Cell Metabol 2011;14: 537–544.

8 Najar M, Raicevic G, Fayyad-Kazan H et al. Impact of different mesenchymal stromal cell types on T-cell activation, proliferation and migration. Int Immunopharmacol 2013;15:693– 702.

9 Wang L, Feng Z, Wang X et al. DEGseq: An R package for identifying differentially expressed genes from RNA-seq data. Bioin-formatics 2010;26:136–138.

10 Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res 2002;30:207–210. 11 Huang DW, Sherman BT, Lempicki RA. Bioin-formatics enrichment tools: Paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res 2009;37:1–13.

12 Wanet A, Remacle N, Najar M et al. Mitochondrial remodeling in hepatic differen-tiation and dedifferendifferen-tiation. Int J Biochem Cell Biol 2014;54C:174–185.

13 Brand MD, Nicholls DG. Assessing mito-chondrial dysfunction in cells. Biochem J 2011;435:297–312.

14 Chen CT, Shih YR, Kuo TK et al. Coordi-nated changes of mitochondrial biogenesis and antioxidant enzymes during osteogenic differentiation of human mesenchymal stem cells. STEMCELLS2008;26:960–968.

(15)

15 Fabregat I, Moreno-Caceres J, Sanchez A et al. TGF-beta signalling and liver disease. FEBS J 2016;283:2219–2232.

16 Palomaki S, Pietila M, Laitinen S et al. HIF-1alpha is upregulated in human mesenchymal stem cells. STEMCELLS2013;31:1902–1909. 17 Vega RB, Huss JM, Kelly DP. The coacti-vator PGC-1 cooperates with peroxisome proliferator-activated receptor alpha in tran-scriptional control of nuclear genes encoding mitochondrial fatty acid oxidation enzymes. Mol Cell Biol 2000;20:1868–1876.

18 Puigserver P, Wu Z, Park CW et al. A cold-inducible coactivator of nuclear recep-tors linked to adaptive thermogenesis. Cell 1998;92:829–839.

19 Rhee J, Inoue Y, Yoon JC et al. Regulation of hepatic fasting response by PPARgamma coactivator-1alpha (PGC-1): Requirement for hepatocyte nuclear factor 4alpha in gluconeo-genesis. Proc Natl Acad Sci USA 2003;100:4012– 4017.

20 Campbell CT, Kolesar JE, Kaufman BA. Mitochondrial transcription factor A regulates mitochondrial transcription initiation, DNA packaging, and genome copy number. Bio-chim Biophys Acta 2012;1819:921–929. 21 Clay Montier LL, Deng JJ Bai Y. Number matters: Control of mammalian mitochondrial

DNA copy number. J Genet Genom 2009;36: 125–131.

22 Nagaoka M, Duncan SA. Transcriptional control of hepatocyte differentiation. Prog Mol Biol Transl Sci 2010;97:79–101. 23 Charos AE, Reed BD, Raha D et al. A highly integrated and complex PPARGC1A transcription factor binding network in HepG2 cells. Genome Res 2012;22:1668– 1679.

24 Puigserver P, Rhee J, Donovan J et al. Insulin-regulated hepatic gluconeogenesis through FOXO1-PGC-1alpha interaction. Nature 2003;423:550–555.

25 Knutti D, Kaul A, Kralli A. A tissue-speciﬁc coactivator of steroid receptors, iden-tiﬁed in a functional genetic screen. Mol Cell Biol 2000;20:2411–2422.

26 Shitashige M, Hirohashi S, Yamada T. Wnt signaling inside the nucleus. Cancer Sci 2008;99:631–637.

27 Dowell KG, Simons AK, Bai H et al. Novel insights into embryonic stem cell self-renewal revealed through comparative human and mouse systems biology networks. STEMCELLS2014;32:1161–1172.

28 McLin VA, Rankin SA, Zorn AM. Repres-sion of Wnt/beta-catenin signaling in the anterior endoderm is essential for liver and

pancreas development. Development 2007; 134:2207–2217.

29 Chen CT, Hsu SH, Wei YH. Mitochondrial bioenergetic function and metabolic plasticity in stem cell differentiation and cellular reprogramming. Biochim Biophys Acta 2012; 1820:571–576.

30 Kawakami Y, Tsuda M, Takahashi S et al. Transcriptional coactivator PGC-1alpha regulates chondrogenesis via association with Sox9. Proc Natl Acad Sci USA 2005;102:2414–2419. 31 Lustig Y, Ruas JL, Estall JL et al. Separa-tion of the gluconeogenic and mitochondrial functions of PGC-1{alpha} through S6 kinase. Genes Dev 2011;25:1232–1244.

32 Fernandez-Marcos PJ, Auwerx J. Regula-tion of PGC-1alpha, a nodal regulator of mitochondrial biogenesis. Am J Clin Nutr 2011;93:884S–8890.

33 Yoshida Y, Shimomura T, Sakabe T et al. A role of Wnt/beta-catenin signals in hepatic fate speciﬁcation of human umbilical cord blood-derived mesenchymal stem cells. Am J Physiol Gastroint Liver Physiol 2007;293: G1089–G1098.

34 Kang S, Bajnok L, Longo KA et al. Effects of Wnt signaling on brown adipocyte differentiation and metabolism mediated by PGC-1alpha. Mol Cell Biol 2005;25:1272–1282.